Individuals who use MA were recruited from 7 government-operated drug rehabilitation centres in Anhui province, China. The eligibility screening criterion for MA as the main drug choice was having used before entering rehabilitation. Written informed consent was obtained. Field collective measurements (approximately 60 people per group) were conducted using a standardized structured questionnaire which included questions about socio-demographics, drug-use behaviours, childhood abuse or neglect, and depression. The participants completed the survey independently or were guided face-to-face by trained interviewers. The research design and methods were approved by the Institutional Review Board of Anhui University of Chinese Medicine. In this study, 1,256 individuals who use MA were recruited, and 1,173 individuals who use MA (aged 18-56) completed the survey; the valid completion rate was 93.4%.

Demographic information on age, gender, educational status, marital status, occupation and past diagnosis of psychiatric disorders or mental disorders were collected. Frequency of MA use was measured with the question: “During the 1 month (i.e., 30 days) prior to entering the drug rehabilitation centre, how many days did you use MA?” In addition to MA, the use of other illicit drugs, such as heroin, morphine, marijuana, ecstasy, ketamine, magu (a mixture of MA and caffeine), and any other addictive drugs named by participants was measured. Years (age) of first drug use and years of last drug use were recorded. Duration of drug use was calculated through first and last drug use. The history of tobacco and wine use were also recorded.

The childhood abuse or neglect of participants was assessed using the short form of the Childhood Trauma Questionnaire (CTQ-SF; Bernstein & Fink, 1998; Bernstein et al., 2003). Contains five subscales (Emotional abuse, Physical abuse, Sexual abuse, Emotional neglect, and Physical neglect) assessing the history of childhood trauma before the age of 16. ‘Moderate/severe’ cut-offs of the CTQ-SF were used to create classification: emotional abuse score ≥13, physical abuse score ≥10, sexual abuse score ≥8, emotional neglect score ≥15 and physical neglect score ≥10 (0 = no/low abuse or neglect; 1 = the presence of moderate/severe abuse or neglect for at least one of the five subscales of CTQ-SF). ‘Moderate/severe’ cut-offs were shown to be good predictors of depression, PTSD, and suicidality (Carlier, Hovens, Streevelaar, van Rood, & Van, 2016). The Chinese version of CTQ-SF was translated and validated by Zhao, Zhang, and Li (2004)). In the current sample, McDonald’s omega was acceptable for the total score (ω = .80) as well as the subscales: Emotional abuse (ω = .70), Physical abuse (ω = .82), Sexual abuse (ω = .78), Emotional neglect (ω = .85), and Physical neglect (ω = .70).

The depression of participants was assessed using the Beck Depression Inventory-II (BDI-II; Beck, Steer, & Brown, 1996). It is a 21-item self-report questionnaire with acceptable reliability and validity in measuring depressive symptoms. The total score of BDI-II ranges from 0 to 63, with higher score showing more depressive symptoms. The cut-offs with different degrees of depression are: 0-13 showing no depression, 14-19 showing mild depression, 20-28 showing moderate depression, and 29-63 showing severe depression (Beck et al., 1996). The Beck Depression Inventory-II in its Chinese Version (BDI-II-C, ω = .87 in current study) has good criterion validity, and the aforementioned cut-offs could be used for detecting Chinese adolescent depression (Yang & Xiong, 2016). In our study, we created classifications of depression through a cut-off of 14; that is, there was no depression when BDI-II scores were less than 14 points, and there was depression when BDI-II scores were greater than or equal to 14 points.

Data were analysed using SPSS version 23.0. The prevalence rates and gender differences of childhood abuse or neglect and depression were calculated and compared. Chi-squared tests (χ2) and t tests were used to evaluate the statistical significance of the differences in demographic characteristics, substance use variables and depression symptoms (BDI-II score≥14: depression; BDI-II score<14: no depression) between the classifications of CTQ-SF scores with ‘moderate/severe’ cut-offs. A Chi-square test and crude odds ratios estimated the relationships between depression and childhood abuse or neglect. Four multiple regressions were employed to predict depression using socio-demographic characteristics, substance use variables, childhood abuse or neglect. Educational status and marital status with more than two levels were dummy-coded. To not rule out potentially important correlates, significance variables were screened using a more liberal p-value (p < .10) (e.g., Vogel, Ramo, & Rubinstein, 2018), and then the screened significance variables in the first three multiple regressions together predicted depression in fourth step (p <  .05). Last, we created a classification of duration of drug use (<10 vs. ≥10 years) and explored the interactions to determine whether childhood abuse or neglect moderated the association between either gender or duration of drug use and depression in separate interactions.

In the current study, 1,173 (762, 65% men and 411, 35% women) participants completed all CTQ-SF and BDI-II assessments and other questions. The prevalence of childhood abuse or neglect and depression was high (see Table 1), with 44.6% (n = 523) of this sample reporting moderate/severe childhood abuse or neglect (the presence of moderate/severe abuse or neglect for at least one of the five subscales of CTQ-SF) and 56.9% (n = 668) were diagnosed with mild to severe depression according to criteria of BDI-II scores ≥14 points. We performed a Chi-Square test to examine gender differences in childhood abuse or neglect and depression. The findings revealed that moderate/severe emotional abuse and neglect of women was much greater than that of men (p < .05, p <  .01), but sexual abuse was the opposite (p <  .05). Moderate/severe physical abuse or neglect of the men who use MA was no different from that of the women who use MA (p >  .05), and total childhood abuse or neglect was no difference between men and women (p >  .05). In addition, there was a difference between men and women who use MA in depression symptoms (p < .001).

Table 2 shows socio-demographic, substance use and depression symptom characteristics and the statistically significant differences between individuals who use MA with no/low and moderate/severe childhood abuse or neglect. The final sample (N = 1,173) ranged in age from 18 to 56 years (M = 31.3, SD = 6.8) and women who use MA comprised 35% (n = 411). Education status was mainly junior middle school or below (79.0%). The participants were unmarried or widowed, divorced, separated (69.2%) or currently unemployed (48.8%); 1.5% had a past diagnosis of psychiatric disorders or mental disorders, and 30.3% reported a history of other drug use in addition to MA. The average age at first drug use was 23.9 years and the average duration of drug use was 6.74 years. Cigarette use (93.7%) and alcohol use (64.7%) history were high in participants.

Individuals who use MA with no/low and moderate/severe childhood abuse or neglect showed statistically significant differences in age, education status, marital status, past diagnosis of psychiatric disorders or mental disorders, alcohol use, age at first drug use, drug use frequency and depression symptom. Specifically, mean age and age at first drug use was younger and drug use frequency was higher in the moderate/severe group. In addition, the participants with moderate/severe childhood abuse or neglect reported more depression.

Table 3 shows the Chi-square test and the crude and adjusted odds ratios (ORs) for the relationships among each of five categories of childhood abuse or neglect and depression. Except for physical and emotional abuse, analyses revealed that three categories of childhood abuse or neglect were significantly associated with increased risk of depression.

The results of multivariable regression analyses of depression scores by childhood abuse or neglect, socio-demographic and substance use variables are respectively presented in Table 4. In Step 1, we entered socio-demographic data (gender, age, occupation, educational status, marital status, past diagnosis of psychiatric disorders or mental disorders) and variables of gender, primary school or illiterate and past diagnosis of psychiatric disorders or mental disorders were screened to statistical significance variables (p <  .10). In Step 2, substance use variables were entered and variables of duration of drug use and other drug use were screened to statistical significance variables (p <  .10). In Step 3, variables of emotional abuse, sexual abuse and emotional neglect were screened to statistical significance variables (p <  .10). Lastly, we entered screening significance variables together in three multiple regressions to predict depression scores (all variables evaluated at p <  .05). The findings revealed that depression scores were significantly predicted by gender, duration of drug use, sexual abuse and emotional neglect (model R2 = .08, p < .001).

In addition, based on the results of the multiple regression analyses, we further explored the interactions to determine whether childhood abuse or neglect moderated the association between either gender or duration of drug use and depression. The interaction between emotional neglect and duration of drug use on lifetime depression was statistically significant (p < .01), showing that individuals exposed to moderate/severe emotional neglect had significantly higher depression scores with longer duration of drug use. A simple slope analysis was used to analyse the moderating effect of emotional neglect. Using duration of drug use (<10 vs. ≥10 years) as a predictor and depression score as a dependent variable, linear regression analysis was carried out. The results showed that the standardized beta of duration of drug use was 0.09 (t = 2.53, p = .01) in the no/low emotional neglect group and was 0.25 (t = 4.61, p < .001) in the moderate/severe emotional neglect group (figure not shown). The interaction between sexual abuse and duration of drug use or gender on lifetime depression was not statistically significant, and the interaction between emotional neglect and gender on lifetime depression was also not statistically significant (figure not shown).